Biological and Environmental Research - Earth and Environmental System Sciences
Earth and Environmental System Modeling
19 September 2020

Characterization of Long-Period Return Values of Extreme Daily Temperature and Precipitation in the CMIP6 Models: Part 1, Model Evaluation


Using a non-stationary Generalized Extreme Value statistical method, we calculate selected extreme daily temperature and precipitation indices and their 20 year return values from the CMIP5 and CMIP6 historically forced climate models. We evaluate model performance of these indices and their return values in replicating similar quantities calculated from gridded land based daily observations. We find that at their standard resolutions, there are no meaningful differences between the two generations of models in their quality of simulated extreme daily temperature and precipitation. This extreme value evaluation framework is currently being incorporated into the PCMDI model intercomparison products.


Our analysis reveals that no single CMIP5 nor CMIP6 model stands out as distinctly superior across either temperature or precipitation extremes. The range of model performance in simulating temperature extremes is comparable between the two generations of climate models and little difference in the performance of multi-model average simulations of annual average temperature and precipitation extremes or their 20-year return values. This is part 1 of a two-part study, the other one examining future climate change. Taken together, regarding extreme weather we conclude that the CMIP5 and CMIP6 models are indistinguishable, in large part because the model resolution in the standard historical configuration is unchanged.


A classical evaluation extreme temperature and precipitation of CMIP6 and CMIP5 historical simulations finds no significant differences between recent generations of climate models.

Michael Wehner
Lawrence Berkeley National Laboratory